一种基于信息熵和密度的K-means算法的改进

An Improvement of K-means Algorithm Based on Information Entropy and Density

影响K-means聚类算法的因素主要有聚类个数、初始聚类中心、异常点、相似性度量和聚类评价准则五个方面。本文通过利用信息熵确定属性的权重,从而对欧氏距离进行加权处理,将孤立点从数据集中取出,从而更好得选出聚类中心,然后利用加权欧氏距离公式对数据集进行相应的聚类。实验结果表明,基于信息熵和密度的K-means聚类算法聚类结果更精确。

The factors affecting the K-means clustering algorithm mainly include five aspects:cluster number,initial cluster center,outlier,similarity measure and cluster evaluation criteria.This paper uses the information entropy to determine the weight of the attribute,thus the Euclidean distance is weighted,and the isolated points are taken out from the data set,so that the cluster center is better selected.Then,the data set is clustered by the weighted Euclidean distance formula.The experimental results show that the K-means clustering algorithm based on information entropy and density is more accurate.